CN109543893A - Heterogeneous Information cyberrelationship prediction technique, readable storage medium storing program for executing and terminal - Google Patents
Heterogeneous Information cyberrelationship prediction technique, readable storage medium storing program for executing and terminal Download PDFInfo
- Publication number
- CN109543893A CN109543893A CN201811358462.5A CN201811358462A CN109543893A CN 109543893 A CN109543893 A CN 109543893A CN 201811358462 A CN201811358462 A CN 201811358462A CN 109543893 A CN109543893 A CN 109543893A
- Authority
- CN
- China
- Prior art keywords
- unmarked
- cluster
- positive
- example set
- cyberrelationship
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Abstract
A kind of Heterogeneous Information cyberrelationship prediction technique, readable storage medium storing program for executing and terminal, which comprises obtain the destination node being originally inputted to set；The destination node pre-processes set, obtains corresponding positive example set and unmarked example set；Based on obtained positive example set, corresponding unmarked example is extracted from the unmarked example set, forms corresponding reliable counter-example set；Corresponding Heterogeneous Information cyberrelationship prediction model is obtained using the positive example set and obtained reliable counter-example set training；The unknown relation between nodes to be predicted is predicted using the Heterogeneous Information cyberrelationship prediction model that training obtains.The accuracy of Heterogeneous Information nodes Relationship Prediction can be improved in above-mentioned scheme.
Description
Technical field
The invention belongs to data analysis technique field, more particularly to a kind of Heterogeneous Information cyberrelationship prediction technique, can
Read storage medium and terminal.
Background technique
With the rapid development of science and technology, the social mode increasingly diversification of people, various complex networks
Thus it is born.From ant colony structure to social intercourse system, from nervous system to the ecosystem, from traffic system to electric system etc. reality
Complication system in the world can topology be approximately complex network structures.Object abstract representation in complication system is in network
Node, link of the interactive relation abstract representation between node between object.In complex network research, link prediction is huge because of its
Big application value is widely paid close attention to by researcher.
Currently, the research object of most of link prediction is the complex network of homogeneity, i.e., node and links category in network
Type is single.However, real complex network is the network of isomery mostly, there are dependences complicated between a plurality of types of nodes and node
Relationship.Homogeneous network substantially be heterogeneous network a homogeneity section, therefore only research homogeneous network can lose it is important
Information.For example, reality social networks in not only exist user node and indicate friends link, further include log,
The nodes of the types such as word, position and timestamp and indicate log and the linking of inclusion relation, log and place between term node
Between register the link etc. of relationship；Node in medical network has the types such as patient, doctor, disease, drug and hospital site.This
A little information have potential influence for the prediction of Object linking.In Heterogeneous Information network, the relationship of node pair can use one
Item is directly linked to indicate, can also be indicated by the path of a mixing multiple types node and link.Therefore, to be predicted
Target may be simple link, it is more likely that it is several link composition relationship.In this way, link forecasting problem just extend in order to
Relationship Prediction problem.
Summary of the invention
Present invention solves the technical problem that being how to improve the accuracy of Heterogeneous Information nodes Relationship Prediction.
In order to achieve the above object, the present invention provides a kind of Heterogeneous Information cyberrelationship prediction technique, which comprises
The destination node being originally inputted is obtained to set；
The destination node pre-processes set, obtains corresponding positive example set and unmarked example set；
Based on obtained positive example set, corresponding unmarked example is extracted from the unmarked example set, composition corresponds to
Reliable counter-example set；
Corresponding Heterogeneous Information cyberrelationship is obtained using the positive example set and obtained reliable counter-example set training
Prediction model；
The Heterogeneous Information cyberrelationship prediction model obtained using training is to the non-MS between nodes to be predicted
System is predicted.
It is optionally, described to pre-process the destination node to set, comprising:
Given network is constructed using destination node type as first set of paths of starting point；
It is special to the number of path in corresponding every member path and random walk to each node in set to calculate destination node
Sign；
Using each node to the number of path and random walk feature construction node pair in every first path for constituting the node
Corresponding example forms example collection；
Using in the example collection there are the node of relationship by objective (RBO) to corresponding example as positive example, there will be no targets to close
The node of system, as unmarked example, obtains the positive example set and unmarked example set to corresponding example.
Optionally, described to be based on obtained positive example set, it is extracted from the unmarked example set corresponding unmarked
Example forms corresponding reliable counter-example set, comprising:
It clusters respectively to the positive example set and unmarked example set, obtains the corresponding positive cluster in part and part is not marked
Remember cluster；
Based on the exemplary feature in the unmarked cluster of the positive cluster in the part and part, the positive cluster in each part is calculated to locally not
Mark the distance between cluster；
Each positive cluster in the part is voted to the unmarked cluster in part for being greater than preset distance threshold with its distance, and
Using the unmarked example in the unmarked cluster in preset quantity part that ballot sum is arranged in front as counter-example, obtain described reliable anti-
Example set.
Optionally, the quantity between the unmarked cluster of the positive cluster in the part and part meets following relationship:
Wherein, N indicates the number of the positive cluster in part, and K indicates the number of the unmarked cluster in part, | U | it indicates not
Exemplary number in example set is marked, | P | indicate exemplary number in positive set.
It is optionally, described to cluster to the positive example set and unmarked example set, comprising:
K-means clustering algorithms are respectively adopted to cluster to the positive example set and unmarked example set.
Optionally, using the distance between following each positive cluster in part of formula calculating to the unmarked cluster in part:
And:
Wherein, d (LPi, ULCj) indicate the positive cluster LP in partiTo the unmarked cluster ULC in partjThe distance between,Expression office
The positive cluster LP in portioniIn example, xiIndicate the positive cluster LP in partiIn exemplary ith feature,Indicate the unmarked cluster in part
ULCjIn example, x 'iIndicate the unmarked cluster ULC in partjIn exemplary ith feature, min () indicate minimum operation, m
Indicate the number of feature in example.
Optionally, described to obtain corresponding isomery with obtained reliable counter-example set training using the positive example set and believe
Cease cyberrelationship prediction model, comprising:
It is "+1 " by each example markup in the positive example set, is by each example markup in reliable counter-example set
" -1 " constitutes corresponding training set；
The training set is inputted in preset Naive Bayes Classifier and is trained, the Heterogeneous Information network is obtained
Relationship Prediction model.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer instruction, described
The step of computer instruction executes Heterogeneous Information cyberrelationship prediction technique described in any of the above embodiments when running.
The embodiment of the invention also provides a kind of terminal, including memory and processor, energy is stored on the memory
Enough computer instructions run on the processor, the processor execute any of the above-described when running the computer instruction
The step of described Heterogeneous Information cyberrelationship prediction technique.
Compared with prior art, the invention has the benefit that
Above-mentioned scheme, by be based on obtained positive example set, extracted from the unmarked example set it is corresponding not
Example is marked, corresponding reliable counter-example set is formed, and trained using the positive example set and obtained reliable counter-example set
To corresponding Heterogeneous Information cyberrelationship prediction model, the confidence level of the counter-example in counter-example set can be improved, therefore can be improved
The accuracy for the Heterogeneous Information cyberrelationship prediction model that training obtains, and then the accuracy of node relationships prediction can be improved.
Detailed description of the invention
In order to more clearly explain the technical solutions in the embodiments of the present application, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, the drawings in the following description are only some examples of the present application, for
For those of ordinary skill in the art, without any creative labor, it can also be obtained according to these attached drawings
His attached drawing.
Fig. 1 is a kind of flow diagram based on Heterogeneous Information cyberrelationship prediction technique of the embodiment of the present invention:
Fig. 2 is a kind of structural schematic diagram based on Heterogeneous Information cyberrelationship prediction meanss of the embodiment of the present invention.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It is based on
Embodiment in the application, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall in the protection scope of this application.Related directionality instruction in the embodiment of the present invention (such as upper and lower, left and right,
It is forward and backward etc.) it is only used for the relative positional relationship explained under a certain particular pose (as shown in the picture) between each component, movement feelings
Condition etc., if the particular pose changes, directionality instruction is also correspondingly changed correspondingly.
As stated in the background art, the existing complex network link/Relationship Prediction method of the prior art mostly uses greatly supervision to learn
Frame is practised, is to label by there are Object linking/relationship nodes in network, from Object linking/relationship section is not present
Point centering is all or random selection a part marks and is, and positive example and the counter-example instruction for needing largely to mark in training process
Practice data to improve the precision of prediction of classifier.But these may will form target in future labeled as the node of counter-example
Link/relationship, therefore it is not necessarily believable counter-example.And complex network general data is larger, to consider to pass through sampling
Big data problem is become into small data problem, negative data is extracted in a random way and is likely to reduce the prediction mould trained
The performance of type.Meanwhile positive example sample (there are Object linking/relationship node to) with unmarked sample (temporarily without Object linking/pass
The node of system to) quantity it is extremely uneven, there is a large amount of unmarked sample, how therefrom to select representative and believable
Counter-example is a good problem to study.
Technical solution of the present invention is extracted from the unmarked example set and is corresponded to by being based on obtained positive example set
Unmarked example, form corresponding reliable counter-example set, and using the positive example set and obtained reliable counter-example set instruction
Corresponding Heterogeneous Information cyberrelationship prediction model is got, can be improved the confidence level of the counter-example in counter-example set, therefore can be with
The accuracy for the Heterogeneous Information cyberrelationship prediction model that training obtains is improved, and then the accurate of node relationships prediction can be improved
Property.
It is understandable to enable above-mentioned purpose of the invention, feature and beneficial effect to become apparent, with reference to the accompanying drawing to this
The specific embodiment of invention is described in detail.
Fig. 1 is a kind of flow diagram based on Heterogeneous Information cyberrelationship prediction technique of the embodiment of the present invention.Referring to
A kind of Fig. 1, item recommendation method based on predicted value filling, can specifically include following step:
Step S101: the destination node being originally inputted is obtained to set.
Step S102: the destination node pre-processes set, obtains corresponding positive example set and unmarked example
Set.
In specific implementation, to it is acquired be originally inputted destination node set pre-processed when, be primarily based on
Given network is constructed using destination node type as first set of paths of starting point.Then, according to being constructed with target section
Vertex type is first set of paths of starting point, calculates each destination node to the number of path and random trip in corresponding every first path
Feature is walked, and using the number of path in every member path of each node pair and random walk feature as in the example of the node pair
Element obtains each node to corresponding example, to form corresponding example collection.After example collection formation, pass through
Corresponding example is added in positive example set as positive example using there are the nodes of relationship by objective (RBO) in the example collection, and will not
There are the nodes of relationship by objective (RBO), and corresponding example to be added in unmarked example set as unmarked example, to finally obtain described
Positive example set and unmarked example set.
Step S103: being based on obtained positive example set, extract corresponding unmarked example from the unmarked example set,
Form corresponding reliable counter-example set.
In the preset implementation, since the flag data in PU study only has positive example, how by these positive examples from
It is critically important that reliable counter-example data are extracted in unmarked set.
In an embodiment of the present invention, use clustering algorithms first, such as K-means clustering algorithms, respectively to it is described just
Example set and unmarked example set cluster, and obtain the corresponding positive cluster in N number of part and the K unmarked cluster in part.Wherein, gained
To the positive cluster in N number of part and the K unmarked cluster in part between quantity meet following relationship:
Wherein, N indicates the number of the positive cluster in part, and K indicates the number of the unmarked cluster in part, | U | indicate unmarked example set
In exemplary number, | P | indicate positive example set in exemplary number.
After obtaining the positive cluster in N number of part and the K unmarked cluster in part, the equidistant calculation method meter of Euclidean distance is being used
The positive cluster in each part is calculated to locally the distance between unmarked cluster.Wherein, in an embodiment of the present invention, using following formula
The positive cluster in each part is calculated to locally the distance between unmarked cluster:
And:
Wherein, d (LPi, ULCj) indicate the positive cluster LP in partiTo the unmarked cluster ULC in partjThe distance between,Expression office
The positive cluster LP in portioniIn example, xiIndicate the positive cluster LP in partiIn exemplary ith feature,Indicate the unmarked cluster in part
ULCjIn example, x 'iIndicate the unmarked cluster ULC in partjIn exemplary ith feature, min () indicate minimum operation, m
Indicate the number of feature in example.
Be calculated the positive cluster in each part between the unmarked cluster in part apart from when, by by calculated distance
It is compared with preset distance threshold, it will be unmarked greater than the part of preset distance threshold with the distance between the positive cluster in part
Cluster as with the positive cluster in part apart from the unmarked cluster in farther away part, from the positive cluster in each part respectively to it apart from farther away part
Unmarked cluster is voted respectively.Wherein, the positive cluster in each part is to the poll thrown with it apart from the unmarked cluster in farther away part
It is identical.When the ballot closes, count the aggregate votes of the unmarked cluster in each part, and will the unmarked cluster in part according to aggregate votes from height
It is arranged to low sequence, is added the unmarked example in the unmarked cluster in the part for arranging first preset quantity as counter-example
The reliable counter-example set finally obtains the reliable counter-example set.
Step S104: corresponding Heterogeneous Information is obtained using the positive example set and obtained reliable counter-example set training
Cyberrelationship prediction model.
In specific implementation, when obtaining the positive example set and the reliable counter-example set, using obtaining the positive example
Set and the reliable counter-example set are trained, and corresponding Heterogeneous Information cyberrelationship prediction model can be obtained.Specifically,
It can be first "+1 " by each example markup in the positive example set, be by each example markup in reliable counter-example set
" -1 " constitutes corresponding training set, then the training set is inputted in preset Naive Bayes Classifier and is trained,
Obtain the Heterogeneous Information cyberrelationship prediction model.In order to further increase to obtain the Heterogeneous Information cyberrelationship prediction mould
The accuracy of type can be assessed with performance of the test set to obtained Heterogeneous Information cyberrelationship prediction model, and selection is most
Excellent model parameter, so that obtaining the Relationship Prediction that Heterogeneous Information cyberrelationship prediction model greatly is optimal after test
Energy.
Step S105: the Heterogeneous Information cyberrelationship prediction model obtained using training is between nodes to be predicted
Unknown relation predicted.
In specific implementation, it when the Heterogeneous Information cyberrelationship prediction model that training obtains, can use acquired
Heterogeneous Information cyberrelationship prediction model the unknown portions of current network are predicted, i.e., in current network be not connected with
Unknown connection relationship between destination node pair predicts, the new network after being predicted.
The above-mentioned Heterogeneous Information cyberrelationship prediction technique in the embodiment of the present invention is described in detail, and below will
The above-mentioned corresponding device of method is introduced.
Fig. 2 shows the structural schematic diagrams of one of embodiment of the present invention Heterogeneous Information cyberrelationship prediction meanss.Ginseng
See Fig. 2, a kind of Heterogeneous Information cyberrelationship prediction meanss 20 may include set acquiring unit 201, set pretreatment unit
202, gather construction unit 203, model training unit 204 and Relationship Prediction unit 205, in which:
The set acquiring unit 201, suitable for obtaining the destination node being originally inputted to set.
The set pretreatment unit 202, suitable for the destination node pre-processes set, obtain it is corresponding just
Example set and unmarked example set.
The set construction unit 203 is suitable for being based on obtained positive example set, extract from the unmarked example set
Corresponding unmarked example forms corresponding reliable counter-example set.
The model training unit 204, trained suitable for the use positive example set and obtained reliable counter-example set
To corresponding Heterogeneous Information cyberrelationship prediction model.
The Relationship Prediction unit 205, the Heterogeneous Information cyberrelationship prediction model suitable for being obtained using training are treated pre-
The unknown relation surveyed between nodes is predicted.
In specific implementation, the pretreatment unit 202, suitable for being with destination node type to given network construction
First set of paths of starting point；Calculate destination node to each node in set to the number of path in corresponding every first path and
Random walk feature；Using each node to the number of path and random walk feature construction section in every first path for constituting the node
Point forms example collection to corresponding example；Corresponding example is made by there are the nodes of relationship by objective (RBO) in the example collection
For positive example, is obtained by the positive example set and is not marked as unmarked example for corresponding example there will be no the node of relationship by objective (RBO)
Remember example set.
In specific implementation, the set construction unit 203 is suitable for respectively to the positive example set and unmarked example set
It clusters, obtains the unmarked cluster of the corresponding positive cluster in part and part；Based in the unmarked cluster of the positive cluster in the part and part
Exemplary feature calculates the positive cluster in each part to locally the distance between unmarked cluster；Each positive cluster Xiang Yuqi in the part away from
The preset quantity office voted from the unmarked cluster in part for being greater than preset distance threshold, and ballot sum is arranged in front
Unmarked example in the unmarked cluster in portion obtains the reliable counter-example set as counter-example.In an embodiment of the present invention, the office
Quantity between the unmarked cluster of the positive cluster in portion and part meets following relationship:
Wherein, N indicates the number of the positive cluster in part, and K indicates the number of the unmarked cluster in part, | U | it indicates not
Exemplary number in example set is marked, | P | indicate exemplary number in positive example set.
In an embodiment of the present invention, the set construction unit 203, suitable for K-means clustering algorithms pair are respectively adopted
The positive example set and unmarked example set cluster.
In an embodiment of the present invention, the set construction unit 203, suitable for calculating each part using following formula
Positive cluster is to locally the distance between unmarked cluster:
And:
Wherein, d (LPi, ULCj) indicate the positive cluster LP in partiTo the unmarked cluster ULC in partjThe distance between,Expression office
The positive cluster LP in portioniIn example, xiIndicate the positive cluster LP in partiIn exemplary ith feature,Indicate the unmarked cluster in part
ULCjIn example, x 'iIndicate the unmarked cluster ULC in partjIn exemplary ith feature, min () indicate minimum operation, m
Indicate the number of feature in example.
In specific implementation, the model training unit 204, suitable for being by each example markup in the positive example set
Each example markup in reliable counter-example set is " -1 ", constitutes corresponding training set by "+1 "；Training set input is pre-
If Naive Bayes Classifier in be trained, obtain the Heterogeneous Information cyberrelationship prediction model.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer instruction, described
The step of Heterogeneous Information cyberrelationship prediction technique is executed when computer instruction is run.Wherein, the Heterogeneous Information
Cyberrelationship prediction technique refers to the introduction of preceding sections, repeats no more.
The embodiment of the invention also provides a kind of terminal, including memory and processor, energy is stored on the memory
Enough computer instructions run on the processor, the processor execute the isomery when running the computer instruction
The step of information network Relationship Prediction method.Wherein, the Heterogeneous Information cyberrelationship prediction technique refers to preceding sections
Introduction, repeat no more.
Using the above scheme in the embodiment of the present invention, by being based on obtained positive example set, from the unmarked example
It extracts corresponding unmarked example in set, forms corresponding reliable counter-example set, and using the positive example set and obtained
Reliable counter-example set training obtains corresponding Heterogeneous Information cyberrelationship prediction model, and the counter-example in counter-example set can be improved
Confidence level, therefore the accuracy for the Heterogeneous Information cyberrelationship prediction model that training obtains can be improved, and then node can be improved
The accuracy of Relationship Prediction.
The basic principles, main features and advantages of the present invention have been shown and described above.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this
The principle of invention, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, the present invention
Claimed range is delineated by the appended claims, the specification and equivalents thereof from the appended claims.
Claims (9)
1. a kind of Heterogeneous Information cyberrelationship prediction technique characterized by comprising
The destination node being originally inputted is obtained to set；
The destination node pre-processes set, obtains corresponding positive example set and unmarked example set；
Based on obtained positive example set, corresponding unmarked example is extracted from the unmarked example set, composition is corresponding can
By counter-example set；
Corresponding Heterogeneous Information cyberrelationship prediction is obtained using the positive example set and obtained reliable counter-example set training
Model；
Using the obtained Heterogeneous Information cyberrelationship prediction model of training to the unknown relation between nodes to be predicted into
Row prediction.
2. Heterogeneous Information cyberrelationship prediction technique according to claim 1, which is characterized in that described by the target section
Point pre-processes set, comprising:
Given network is constructed using destination node type as first set of paths of starting point；
Calculate number of path and random walk feature of the destination node to each node in set to corresponding every first path；
Using each node to the number of path and random walk feature construction node in every first path for constituting the node to correspondence
Example, formed example collection；
Using in the example collection there are the node of relationship by objective (RBO) to corresponding example as positive example, there will be no relationship by objective (RBO)
Node, as unmarked example, obtains the positive example set and unmarked example set to corresponding example.
3. Heterogeneous Information cyberrelationship prediction technique according to claim 1, which is characterized in that described based on obtained
Positive example set extracts corresponding unmarked example from the unmarked example set, forms corresponding reliable counter-example set, comprising:
It clusters respectively to the positive example set and unmarked example set, obtains the corresponding positive cluster in part and part is unmarked
Cluster；
Based on the exemplary feature in the unmarked cluster of the positive cluster in the part and part, it is unmarked to part to calculate the positive cluster in each part
The distance between cluster；
Each positive cluster in the part is voted to the unmarked cluster in part for being greater than preset distance threshold with its distance, and will be thrown
The unmarked example in the unmarked cluster in preset quantity part that ticket sum is arranged in front obtains the reliable counter-example collection as counter-example
It closes.
4. Heterogeneous Information cyberrelationship prediction technique according to claim 3, which is characterized in that the positive cluster drawn game in part
Quantity between the unmarked cluster in portion meets following relationship:
Wherein, N indicates the number of the positive cluster in part, and K indicates the number of the unmarked cluster in part, | U | it indicates to show in unmarked example set
The number of example, | P | indicate exemplary number in positive example set.
5. Heterogeneous Information cyberrelationship prediction technique according to claim 3, which is characterized in that described to the positive example collection
It closes and unmarked example set clusters, comprising:
K-means clustering algorithms are respectively adopted to cluster to the positive example set and unmarked example set.
6. Heterogeneous Information cyberrelationship prediction technique according to claim 3, which is characterized in that use following formula meter
The positive cluster in each part is calculated to locally the distance between unmarked cluster:
And:
Wherein, d (LPi, ULCj) indicate the positive cluster LP in partiTo the unmarked cluster ULC in partjThe distance between,Indicate part just
Cluster LPiIn example, xiIndicate the positive cluster LP in partiIn exemplary ith feature,Indicate the unmarked cluster ULC in partjIn
Example, x 'iIndicate the unmarked cluster ULC in partjIn exemplary ith feature, min () indicates minimum operation, and m shows
The number of feature in example.
7. Heterogeneous Information cyberrelationship prediction technique according to claim 3, which is characterized in that described to use the positive example
Set and obtained reliable counter-example set training obtain corresponding Heterogeneous Information cyberrelationship prediction model, comprising:
By each example markup in the positive example set be "+1 ", by each example markup in reliable counter-example set be "-
1 ", constitute corresponding training set；
The training set is inputted in preset Naive Bayes Classifier and is trained, the Heterogeneous Information cyberrelationship is obtained
Prediction model.
8. a kind of computer readable storage medium, is stored thereon with computer instruction, which is characterized in that the computer instruction fortune
Perform claim requires the step of 1 to 7 described in any item Heterogeneous Information cyberrelationship prediction techniques when row.
9. a kind of terminal, which is characterized in that including memory and processor, storing on the memory can be in the processing
The computer instruction run on device, perform claim requires described in 1 to 7 any one when the processor runs the computer instruction
Heterogeneous Information cyberrelationship prediction technique the step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811358462.5A CN109543893A (en) | 2018-11-14 | 2018-11-14 | Heterogeneous Information cyberrelationship prediction technique, readable storage medium storing program for executing and terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811358462.5A CN109543893A (en) | 2018-11-14 | 2018-11-14 | Heterogeneous Information cyberrelationship prediction technique, readable storage medium storing program for executing and terminal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109543893A true CN109543893A (en) | 2019-03-29 |
Family
ID=65847372
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811358462.5A Pending CN109543893A (en) | 2018-11-14 | 2018-11-14 | Heterogeneous Information cyberrelationship prediction technique, readable storage medium storing program for executing and terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109543893A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111310822A (en) * | 2020-02-12 | 2020-06-19 | 山西大学 | PU learning and random walk based link prediction method and device |
| CN111414408A (en) * | 2020-03-11 | 2020-07-14 | 成都数融科技有限公司 | Method and device for trusted representation of data based on block chain |
| CN113159357A (en) * | 2020-01-07 | 2021-07-23 | 北京嘀嘀无限科技发展有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104077351B (en) * | 2014-05-26 | 2017-01-25 | 东北师范大学 | Heterogeneous information network based content providing method and system |
| CN106778894A (en) * | 2016-12-29 | 2017-05-31 | 大连理工大学 | A kind of method of author's cooperative relationship prediction in academic Heterogeneous Information network |
| CN107451613A (en) * | 2017-07-31 | 2017-12-08 | 广州市香港科大霍英东研究院 | The semi-supervised learning method and device of Heterogeneous Information network |
-
2018
- 2018-11-14 CN CN201811358462.5A patent/CN109543893A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104077351B (en) * | 2014-05-26 | 2017-01-25 | 东北师范大学 | Heterogeneous information network based content providing method and system |
| CN106778894A (en) * | 2016-12-29 | 2017-05-31 | 大连理工大学 | A kind of method of author's cooperative relationship prediction in academic Heterogeneous Information network |
| CN107451613A (en) * | 2017-07-31 | 2017-12-08 | 广州市香港科大霍英东研究院 | The semi-supervised learning method and device of Heterogeneous Information network |
Non-Patent Citations (1)
| Title |
|---|
| GAO-JING PENG: "A relation prediction method based on PU learning", 《2017 12TH INTERNATIONAL CONFERENCE ON INTELLIGENT SYSTEMS AND KNOWLEDGE ENGINEERING (ISKE)》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113159357A (en) * | 2020-01-07 | 2021-07-23 | 北京嘀嘀无限科技发展有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN113159357B (en) * | 2020-01-07 | 2023-11-24 | 北京嘀嘀无限科技发展有限公司 | Data processing method, device, electronic equipment and computer readable storage medium |
| CN111310822A (en) * | 2020-02-12 | 2020-06-19 | 山西大学 | PU learning and random walk based link prediction method and device |
| CN111310822B (en) * | 2020-02-12 | 2022-09-20 | 山西大学 | PU learning and random walk based link prediction method and device |
| CN111414408A (en) * | 2020-03-11 | 2020-07-14 | 成都数融科技有限公司 | Method and device for trusted representation of data based on block chain |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI766618B (en) | Key point detection method, electronic device and computer readable storage medium | |
| Pan et al. | Evaluating the accessibility of healthcare facilities using an integrated catchment area approach | |
| CN109543893A (en) | Heterogeneous Information cyberrelationship prediction technique, readable storage medium storing program for executing and terminal | |
| CN109523415A (en) | Heterogeneous Information cyberrelationship prediction meanss | |
| Naghibi et al. | Urban growth modeling using cellular automata with multi-temporal remote sensing images calibrated by the artificial bee colony optimization algorithm | |
| Li et al. | MCR-Modified CA–Markov model for the simulation of urban expansion | |
| CN114819589A (en) | Urban space high-quality utilization determination method, system, computer equipment and terminal | |
| Chen et al. | Quantitative evaluation of spatial differentiation for public open spaces in urban built-up areas by assessing SDG 11.7: a case of Deqing County | |
| CN108804870A (en) | Key protein matter recognition methods based on Markov random walks | |
| Gao et al. | Towards travel recommendation interpretability: Disentangling tourist decision-making process via knowledge graph | |
| CN103473416A (en) | Protein-protein interaction model building method and device | |
| Li et al. | Application of clustering algorithms in the location of electric taxi charging stations | |
| Li et al. | An automatic plant leaf stoma detection method based on YOLOv5 | |
| Pei et al. | Self-supervised spatiotemporal clustering of vehicle emissions with graph convolutional network | |
| Shen et al. | GBT: Two-stage transformer framework for non-stationary time series forecasting | |
| Zhou et al. | Fault Diagnosis Method of Smart Meters Based on DBN-CapsNet | |
| Wang et al. | Relationship between urban three-dimensional spatial structure and population distribution: A case study of Kunming’s main urban district, China | |
| CN106779181A (en) | Method is recommended by a kind of medical institutions based on linear regression factor Non-negative Matrix Factorization model | |
| Liu et al. | LUCC simulation based on RF-CNN-LSTM-CA Model with high-quality seed selection iterative algorithm | |
| Guan et al. | Development Process, Quantitative Models, and Future Directions in Driving Analysis of Urban Expansion | |
| Cui et al. | Remote sensing scene graph and knowledge graph matching with parallel walking algorithm | |
| Gul et al. | Hill Climbing-Based Efficient Model for Link Prediction in Undirected Graphs | |
| Tian et al. | Grading evaluation of the structural connectivity of river system networks based on ecological functions, and a case study of the Baiyangdian wetland, China | |
| Cui et al. | A Novel Imputation Model for Missing Concrete Dam Monitoring Data | |
| Oluwasanmi et al. | Multi-head spatiotemporal attention graph convolutional network for traffic prediction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190329 |